Improving the prediction of the fatty acid profile of olive oils by considering statistically relevant harvests and agro-climatic variables.

BACKGROUND

Fatty acids are the major components in extra virgin olive oil, and they are considered as a quality parameter to its authentication. As fraudulent practices are the most important problem in this sector, fast, reliable and cost-effective techniques, such as Raman spectroscopy, have been successfully applied, in combination with chemometrics, to determine the fatty acid profile of oils.

RESULTS

The huge amount of information provided by Raman spectra is reduced in a few orthogonal components of principal component analysis (PCA). The goodness-of-fit of the statistical models including only these PCA factors is considerably increased by introducing dummy variables, associated with the harvest, and some agro-climatic variables (temperature, humidity, wind speed, radiation, precipitation and evapotranspiration). Many of these additional variables are statistically relevant in models using either the global sample or subsamples of Andalusian provinces or olive varieties.

CONCLUSIONS

The regression models using only Raman spectral information are clearly improved by the consideration of harvesting time and agro-climatic data, a useful result as trade standard applying to olive oils limits values for disaggregated fatty acids to authenticate olive oils. Nevertheless, the effect (or the statistical relevance) of these variables depends on the specific type of fatty acid, geographical region (province) or olive variety. © 2019 Society of Chemical Industry.